Adding a Shadow to a Line Chart in ggplot2

Shadowing your ggplot lines.

Line charts with shadows let one feel in control. I’ll use the weekly death statistics from Istanbul to illustrate what I mean.

The data can be accessed by downloading the post25.xlsx file here or visiting this site.

library(tidyverse)

df <- readxl::read_excel("data.xlsx") %>% 
  select(-day,-y2020,-y2021) %>% 
  pivot_longer(!week, names_to = "years", values_to = "values") %>% 
  group_by(week,years) %>% 
  summarise(total = sum(values)) %>% 
  ungroup() %>% 
  pivot_wider(names_from = "years", values_from = "total")

Let’s see the lines before adding a shadow.

df %>% 
  pivot_longer(!week, names_to = "years", values_to = "values") %>% 
  ggplot(aes(x = week, y = values, group = years, color = years)) +
  geom_line() +
  ggthemes::theme_fivethirtyeight() +
  theme(axis.text = element_text(size = 20),
        strip.text = element_text(size = 20),
        legend.title = element_blank(),
        legend.text = element_text(size = 20),
        legend.key.size = unit(1, 'cm'))

First, we’ll create a ribbon by determining the minimum and maximum values by week from deaths between 2015 and 2019.

df2 <- df %>% 
  mutate(
    normal_min = apply(X = .[,2:6], MARGIN = 1, FUN = function(x) min(x)),
    normal_max = apply(X = .[,2:6], MARGIN = 1, FUN = function(x) max(x)),
    normal_mean = apply(X = .[,2:6], MARGIN = 1, FUN = function(x) mean(x))
  )

The following things are easier after the above step. We can add shadowed areas to our lines using geom_ribbon().

lastWeek <- df2 %>% 
  pull(y2022) %>% 
  na.omit() %>% 
  length(.)

ggplot(df2, aes(x = week)) +
  geom_ribbon(
    aes(
      ymin = normal_min,
      ymax = normal_max
    ),
    fill = "dark blue",
    alpha = .5
  ) +
  geom_line(aes(y = normal_mean)) +
  geom_line(aes(y = y2022), color = "dark red", size = 2) +
  geom_vline(xintercept = lastWeek, linetype = "dotdash", size = 1) +
  ggthemes::theme_fivethirtyeight() +
  theme(axis.text = element_text(size = 20),
        axis.title.x = element_text(size = 20)) +
  scale_x_continuous(breaks = c(1,lastWeek,52)) +
  scale_y_continuous(labels = scales::comma) +
  labs(
    title = "Weekly Deaths in Istanbul (from all causes), 2015-2019 vs 2022",
    x = "Weeks"
  )